EVER WANTED TO TURN
SOMEONE INTO A FROG?
Illustrative example. Same expression. More frog.
PICTURES
Create and transform images. Keep your gloriously bad ideas coming.
ILLUSTRATIVE STUDIO EXAMPLE · EDIT · EXPERIMENT · COMPARE
THE IDEAS SHED PRESENTS · ISSUE No. 01
OPEN THE STUDIOIf you could have any superpower, what would it be?
Beta plans: Free $0 · Standard $5/mo · Full $10/mo
Meet Super (ish) Cow. Your sidekick. Mostly here for moral support.
Choose a power. I’ve almost certainly got this.
YOUR IMAGINATION. OUR VERY QUESTIONABLE ASSISTANT.
Something beautiful.
Something ridiculous.
Preferably both.
EVER WANTED TO TURN
SOMEONE INTO A FROG?
Illustrative example. Same expression. More frog.
Create and transform images. Keep your gloriously bad ideas coming.
ILLUSTRATIVE STUDIO EXAMPLE · EDIT · EXPERIMENT · COMPARE
PICTURES ARE ALL VERY WELL.
WHAT IF THEY MOVED?
Press Play. This is the original silent film.
You direct. We make the pictures move. The cow takes most of the credit.
YOUR IDEA · ON SET · PREMIERE FOUNDER PREVIEW
AND NOW…
SOMETHING FOR THE EARS.

Press the record for a short, original cartoon sound effect.
Create voices, sound effects and music. The bar has been set exceptionally low.
SPEAK · MAKE NOISE · COMPOSE FOUNDER PREVIEW
These examples play without starting a generation. Your own creation begins in the studio, with its price and availability shown before you submit.
THREE EXTREMELY GOOD EXCUSES.
For a laugh.
For an idea.
For something that matters.
AN IDEA TO TRY
Use a photo you have permission to edit. Keep the result private or share it with their agreement.
AN IDEA TO TRY
Start with a photo you own. Explore the picture, then develop a story, a short film or a soundtrack around it.
AN IDEA TO TRY
Use an authorised product photo. Check the product and any claims before publishing; finish precise text in your usual design tool.
Open an idea, borrow the brief, then make it your own. Opening a studio keeps your existing photos, prompts and results.
THE PUBLIC LEDGER · HIGHLY QUESTIONABLE ECONOMICS
The Bank of Super (ish) Cow announces another irresponsible fiscal stimulus package.

Checking the Bank’s books…
His pockets contain one moth and a deeply optimistic budget.
COME BACK TOMORROWThe confirmed reset time is below. Your personal coins remain separate.From today’s —-coin shared fund.
YOUR DAILY SHARE
Checking your daily eligibility…
SpentHeld
Time until reset
Create the character or scene in Photos. Download the image you want to use.
Open Photos →Add that image as a starting frame in Movies. Describe one action and one camera move.
Open Movies →Create speech, effects or music in Sound. Download each part separately.
Open Sound →Assemble and mix the files in your video editor. Sound offers a listening preview with your movie; this studio does not export a finished combined film.
Explore the Gallery →Each generation has its own availability, price and permissions. Opening another studio keeps your work in this tab. Download files before leaving.
THE STILL PICTURE DEPARTMENT
Bring a photo. Give it a gloriously bad idea.

THE MOVING PICTURE DEPARTMENT
You direct. We make the pictures move.
The cow has appointed himself executive producer.

“The important bit is
probably the kettle.”
The cow’s explanation is optional. The useful explanation is below.
Use a starting image if you have one. Say what moves, what the camera does, and what should remain still. Start with one clear action. The model creates a short sequence; it can still get movement or continuity wrong.
Check the duration and price before you start. Download your clip, then create sound separately if you need it.
A short prompt about one shot is easier to evaluate than a complete story. Describe camera movement separately from the subject’s action. A starting image can help establish a character and composition, but it does not guarantee identity across every frame.
Make a few purposeful shots, download them, and arrange them in an external video editor. The Sound studio can make voices, effects and music for the sequence.
Video models generate a sequence while trying to maintain relationships between frames. More frames, larger outputs and longer clips usually require more computation. Temporal consistency, hands, physics, text and repeated identities remain model-dependent limitations.
The movie panel names the current model, duration options and estimated provider cost. Availability and budgets are separate from image credits. Sound’s movie pairing is a listening preview; combined audiovisual export is not available here.

Big screen energy.
Small shed budget.
The cameras are nearly rolling.Connecting to the model…
0s elapsedDownload now. Provider files normally expire after about an hour.

“I'll be in my trailer.
It's a wheelbarrow.”
THE SLIGHTLY LOUDER DEPARTMENT
A voice for your cow. A noise for your nonsense.
A soundtrack for whatever happens next.

“The important bit is
probably the kettle.”
The cow’s explanation is optional. The useful explanation is below.
For speech, write the words to be spoken and choose a stock voice. For an effect, describe the sound and its setting. For music, describe a mood, style and instrumentation. Listen before you use the result.
Use words and material you have rights to use. A generated voice is not a recording of a real person.
Speech delivery, pace and pitch can change the feel of the same sentence when the selected model supports them. For effects, describe the source, distance and environment. For music, a focused brief helps more than a list of contradictory genres.
Compare only the models you need. Each requested take is a separate operation. Saved character voices retain stock-voice settings; they do not clone a person.
Speech synthesis maps text and voice settings to audio. Sound-effects and music models generate from descriptive conditioning; their length, output format and parameter support vary. Inspect the actual selected model and cost estimate before submission.
Sound uses its own provider budget. Failed or uncertain attempts may still consume provider compute; the image-wallet refund policy does not replace Sound’s accounting. Pairing audio with a movie starts separate players, so timing can drift and no combined file is exported.

Every great voice
starts with hello.
HEAR IT WITH YOUR MOVIE
Starts this track with the movie. Movie audio is muted. This is a listening preview; download movie and sound separately. Combined export is not available yet.

“I have perfect pitch.
Mostly for selling things.”
A voice for your villain. A theme tune for your entrance. A tiny movie nobody asked for.
Use speech or singing you own or have permission to use.
One paid fal request per click. Pricing varies by tool, video duration and audio length. Check this tool’s current price & API . No automatic paid retries.
THE UNNECESSARILY DRAMATIC REVEAL
Your next “you have to see this”
goes here.
Ready when you are.
DOWNLOAD YOUR CREATIONBeta generation is being connected. You can browse models and sign in; no paid request will start until the service is ready.
Get a fal.ai API keyTHE IDEAS SHED PATENT OFFICE · HIGHLY QUESTIONABLE MACHINERY
Insert a friend. Add a bad idea. Pull the lever.


“The important bit is
probably the kettle.”
The cow’s explanation is optional. The useful explanation is below.
The model uses your photographs and words as guidance to create a new image. Say what should change and what should stay. “Keep the expression” is useful. Download a result you want to keep.
It can invent details or change a face. Check the result, and use photographs you have permission to edit.
Try the same photo and prompt with two models to see how they interpret it. Describe the subject, setting, lighting and the details to preserve. Change one instruction at a time so you can tell what helped.
Comparisons make separate requests and cost more in total. The quote shows the complete selected run before you generate. A longer prompt is not automatically a better prompt.
Image editing models condition generation on text and one or more reference images. Many use diffusion or flow-based sampling; an input image guides synthesis rather than guaranteeing an exact pixel-level edit.
Resolution, reference limits, seeds and exposed parameters vary by model. Open Model & creative controls for the actual provider, model, capabilities and supported settings. Latency includes model loading, queueing, inference and file delivery. Read the live quote; model names alone do not determine a price.
Your choice of photo-editing models.
P-Image and Grok are AI models. Replicate and fal.ai host models; xAI makes Grok and also offers it directly. Choose where to run it:
| AI provider | Models here | Filtering | Best for |
|---|---|---|---|
| Replicate | P-Image, Qwen, FLUX & more | Varies by model. Provider checks remain on. | Low-cost edits and model comparisons |
| fal.ai | Grok, Qwen & FLUX | Relaxed screening can be requested for Grok, but fal or xAI can still refuse. | Grok edits, video, singing & voice tools |
| xAI | Grok Imagine 2.0 | Provider-controlled. No documented off switch. | Grok editing directly from its maker |
Each provider has its own API key and credit. Switching providers does not transfer your balance or remove its rules.
Edit one to six photos into one image for one Mischief Coin. Creative Freedom is our app default: your prompt is sent unchanged, without a custom profanity filter. The model’s hosted checker remains on. Grok on fal.ai is still available as another choice.
This restores our recommended model, provider and published image settings each time you open the page. Your deliberate changes still apply during use.
Choose a model without losing your photos or prompt.
Model details & current pricingUses your existing server-side Replicate token. Choose a model that accepts photos and returns an image. Other providers need their own connection.
One photograph. One prompt. Two very different ways to make a mess.
Use the photo, prompt and consent below. Both receive exactly the same photo and words. Your normal model settings stay as they are.
Checking GPU status…
A PERFECTLY REASONABLE EXAMPLE
“Turn my mate into a frog. Keep the expression.”
TWO TAKES. YOU DECIDE.Same mate. Different worlds.
Illustrative examples, not a benchmark or guaranteed model result. Your real creations will appear here.Sending the idea to your selected provider…
0sUses the displayed image as the reference. Each edit is a new paid generation with your selected model.
Your result stays in this tab until you clear it or leave.
SUPER HERO starts with P-Image Edit on Replicate for photo editing. Grok Imagine 2.0 on fal.ai remains an option. The studio also connects MiniMax H3 Max for video, OmniHuman 1.5 for talking and singing portraits, Wan Effects for visual effects, and named voice and music tools.
Studio tools need your funded fal connection. Qwen editing on Replicate and Grok via direct xAI remain available as explicit choices. Every provider keeps its own rules and can change its service.
Public beta. Registered, verified accounts receive 10 welcome coins; the current price is shown before generation. Recommendations are provisional, not benchmark rankings.
Extraordinary pictures. Very ordinary prices.
MORE COINS. SAME NONSENSE.$5 / month
75 coins each month
25 more than a $5 top-up

$10 / month
160 coins each month
50 more than a $10 top-up
Both memberships renew monthly. Cancel anytime.
Monthly coins reset each billing cycle. Prices in US dollars.
Pay as you go.
No subscription.
Purchased coins never expire.
Choose your level of mischief. No cape surcharge. Image prices are shown before generation. Movies and Sound have separate allowances and costs.

An ordinary English shed. Some deeply unordinary ideas inside.
Step inside. Best with sound.
Ideas, experiments and the occasional glorious mess.
COME AND SEE WHAT'S INSIDE